The neuronal correlates of visual attention are typically studied by using a single electrode and recording the signals of individual neurons. Because the firing of a single neuron varies considerably from trial to trial, responses from dozens of trials must be averaged to obtain precise measurements. But the single-neuron approach is problematic for studies of attention because attention drifts from trial-to-trial in uncontrolled ways that are not captured when data are averaged over trials. A comprehensive understanding of visual attention therefore requires investigation at a finer time scale (seconds or less), in keeping with the time course of natural behaviors. Recent advances in multi-electrode recording can overcome the limitations of studying attention using single-neuron recordings. This approach reveals new facets of the dynamics of attention that previously could not be measured. Using this newfound temporal precision, this study addresses two central questions in visual neuroscience. First, we will determine whether shifts in attention and eye movements are coupled. Second, we will investigate the deployment of attention during natural, unrestricted scanning of a visual display. These two aims will help bridge the gap between our basic research knowledge and translation of that knowledge to clinically viable treatments of attention. The ability to predict and dissociate the neuronal signals for eye movements and visual attention at a milliseconds timescale represents a major step towards this goal. Chronic implants for the online monitoring of aberrant attention-or even for attentional enhancement-depend on the research outlined in this proposal.